gseif's profile - activity

Try this link. It helped me out.

http://stackoverflow.com/questions/22...

Have you tried using the RGB colour space? If you have the image in RGB then only look at channel 3 of the image. Then you would say, if the blue component is greater than a certain threshold then we have detected a blue colour. Anything less than the threshold and its probably a mix of colours.

mask = cv2.inRange(img, lower_magnitude_threshold, 255);

Hope this helps!

It's actually not possible to read in directly grayscale from a webcam. Most webcams will read in by default either YUV or YCrCb. As berak mentioned, 'CV_CAP_PROP_CONVERT_RGB' allows you to set whether or not you want this to be converted to RGB right away. So to get grayscale from a webcam, you'll have to use the 'cvtColor' function like so:

cvtColor(original, gray, CV_BGR2GRAY);

You'll have to do this for every incoming frame. Hope this helps!

Hi,

I have in my code the line. fn_frontFace is a string with the location of the XML Haar classifier.

Ptr<cuda::cascadeclassifier> cascade_gpu = cuda::CascadeClassifier::create(fn_frontFace);

I'm trying to enable face detection with CUDA. But its giving me an error: OpenCV Error: GPU API call NCV Assertion failed in anonymous-namespace::NCVDebugOutputHandler file ....\cascadeclassifier.cpp line 156.

I've tested out other CUDA enabled functions like:

cuda::createBoxFilter() cuda::cvtColor()

and they work just fine. Any ideas on how to solve the error?

Thanks